gp: subst() anomaly

[Michael Somos <somos@grail.cba.csuohio.edu>]

pari-dev,

? E=ellinit([0,0,0,1,0]);
? P=[x,y];
? Q=elladd(E,P,P);
? subst(Q,y^2,x^3+x)
%4 = [0, 0]
? subst(Q[1],y^2,x^3+x)
%5 = (1/4*x^4 - 1/2*x^2 + 1/4)/(x^3 + x)

seems a bit peculiar to me. I can't seem to find a simpler example
of this behavior, although this example is a natural one. Shalom,
Michael